Instructions to use ConvLab/roberta-base-trippy-dst-multiwoz21 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ConvLab/roberta-base-trippy-dst-multiwoz21 with Transformers:
# Load model directly from transformers import AutoTokenizer, TransformerForDST tokenizer = AutoTokenizer.from_pretrained("ConvLab/roberta-base-trippy-dst-multiwoz21") model = TransformerForDST.from_pretrained("ConvLab/roberta-base-trippy-dst-multiwoz21", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Update README.md
Browse files
README.md
CHANGED
|
@@ -35,13 +35,13 @@ python3 run_dst.py \
|
|
| 35 |
--dataset_config=dataset_config/unified_multiwoz21.json \
|
| 36 |
--do_lower_case \
|
| 37 |
--learning_rate=1e-4 \
|
|
|
|
| 38 |
--max_seq_length=180 \
|
| 39 |
--per_gpu_train_batch_size=24 \
|
| 40 |
--per_gpu_eval_batch_size=32 \
|
| 41 |
--output_dir=results \
|
| 42 |
--save_epochs=2 \
|
| 43 |
--eval_all_checkpoints \
|
| 44 |
-
--logging_steps=10 \
|
| 45 |
--warmup_proportion=0.1 \
|
| 46 |
--adam_epsilon=1e-6 \
|
| 47 |
--weight_decay=0.01 \
|
|
|
|
| 35 |
--dataset_config=dataset_config/unified_multiwoz21.json \
|
| 36 |
--do_lower_case \
|
| 37 |
--learning_rate=1e-4 \
|
| 38 |
+
--num_train_epochs=10 \
|
| 39 |
--max_seq_length=180 \
|
| 40 |
--per_gpu_train_batch_size=24 \
|
| 41 |
--per_gpu_eval_batch_size=32 \
|
| 42 |
--output_dir=results \
|
| 43 |
--save_epochs=2 \
|
| 44 |
--eval_all_checkpoints \
|
|
|
|
| 45 |
--warmup_proportion=0.1 \
|
| 46 |
--adam_epsilon=1e-6 \
|
| 47 |
--weight_decay=0.01 \
|